POST A COMMENT

62 Comments

Back to Article

  • Anonymous User - Saturday, October 18, 2003 - link

    BFG Asylum nVidia FX 5950 Ultra 256MB

    Grab it @

    http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&...
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    From a technical / technological aspect I find the Overdrive technology very interesting.

    I equate it to my background in automobile tuning. Carburated engines had a set timing. You set that timing sso the car would not knock and ping. Electronically controlled fuel injection allowed using different timing / fueling at different load point to extract the most from the engine. Same engine could maybe get 10% more power. Now you see very detailed EFI, where the engine computer advances and retards timing dynamically depending on octane, load and many other inputs, always giving near maximum possible output.

    I see this as the beginning of the same trend in graphics cards. It makes sense.
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    This review was done with stock cooling. Those cards get hot real fast.

    Since Overdrive stops overclocking when it gets too hot, I would have like to see some form of ultra cooling on some cards like (at least) water-cooling.

    The way I see it, Overdrive it pointless.

    Reply
  • Anonymous User - Sunday, October 12, 2003 - link

    ahaha @ 57 Reply
  • Anonymous User - Sunday, October 12, 2003 - link

    How long before we can buy a XT watercooled out of the box? I think that's the question remaining now. Reply
  • Anonymous User - Saturday, October 11, 2003 - link

    cause your an idiot Reply
  • Anonymous User - Friday, October 10, 2003 - link

    why i cannot run halo i have a radeon 9000 Reply
  • Anonymous User - Friday, October 10, 2003 - link

    52 - That is exactly what Anand did. They ran a benchmark - recorded the results - then ran another benchmark, and so on. That sounds like they did let it cool every 10 minutes or so. Unlike a real user.

    50 is partially right. The bus speed improvement when OC is enabled would give much better indication of what the XT is doing over some statistically ambivalent fps count.
    Reply
  • Anonymous User - Friday, October 10, 2003 - link

    The HL2 stolen code setback seems to be perfect timing. Nivida has to be loving it, the bastards!

    The only good Dx9 game that should have been out with the Radeon XT would have put a nail in Nvidia.

    Now Nvidia has breathing room. Lets see if they squander it.

    Right now the clear winner for Dx9 is the ATI XT. That is if you are comparing the XT to the 5900 and not some vapour hardware.

    You made the charts, take a look at them. ATI is clearly better at DX9 PS2.0. All the new DX9 games are not going to run better on Nvidia. The benchmarks for HL2 were already shown.

    I don't see why you can't recommend a card at this time?

    If you had 10,000 to bet which card would you pick to be the best dx9 card? If you bet on Nvidia your gonna be 10,000 poorer.

    Reply
  • Anonymous User - Friday, October 10, 2003 - link

    I agree, its much better than the last review. Thanks for the Fps in Tomraider.

    Do you know what the cap is for the XTs auto overclocking?

    Wonder what it would do if you put water cooling on the ATI?
    Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    so... 50 and 51 ... do you play your games for 10 minutes, then turn off the computer to let it cool down so you can get better fps? hell no. you play and play and play till yer done.

    doing all these benchmarks without letting the card cool down (whether it shows the 3.8 full potential or not) is more like what users will experience on their systems. That makes this data more useful to me than anything else they could have done.

    crazy ati fanbois

    Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    I'm with 50. He is right. Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    trying to test a temp related mod without controlling temp in the case is just assinine. This proves absolutely nothing. They themselves said they did not allow cooling between tests so they don't know how the overdrive was working. I would be more impressed with just the change in speed numbers than the fps. That would at least tell us whether the overclocking was used throughout the test and not shut off after the card got hot on the 2nd benchmark. The test had no control basis and therefore all the data is trash. Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    when the 9600 XT is going to be realased, i'm waiting for it Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    In your discussion of image quality, I don't see
    a comment on HALO. Yet on my PC (P4C 2.8,
    Radeon 9800 PRO, Catalyst 3.6) I get clear drawing
    errors. Looking at the sky in any outdoor scene in
    level 5 or 8, it shows just a gaping hole which
    should've been obscured by fog. More subjectively,
    the fog in the swamps at the start/end of level 6
    also doesn't work as well as on the XBOX.

    What are the experiences of people with different
    hardware/drivers?
    Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    I dont see the overdrive tab, is this only visible with 9800pro etc.
    My card is a 9500pro!!
    Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    ... so they will only allow the Mhz, where every card works? Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    maybe we are gonna see unofficial moded versions of these drivers, with the overclocking limit removed :d . ATI just cant allow this difference between the users cards, when one has a "overclocker" card and gets his 10% and the other has something "allergic to speed" card lol and gets nothing. ATI just cant make the speed of the card luck Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    I'm disappointed that Anand didn't try to find out why overdrive didn't produce better figures.

    Overdrive might not overclock as much as some people want, but it should still give more impact than just 0,5fps.
    Since the 9800XT is mainly an overclocked 9800Pro we should be able to accurately estimate the impact of overdrive on the 9800XT.

    I suspect that we would get far better numbers if Anand had used 1024*768 instead of 1600*1200 4xAA.
    Afaik overdrive only overclocks the GPU and not the memory. At 1600*1200 you are more likely to run into a memory bandwith limited situation. (Especially with all those DX8 games)
    In such a situation it is logical that overdrive won't give any large increase in fps.

    The other explanation is that the air termperatuur in the computer case Anand uses was very high and therefor overdrive was not able to overclock much.

    In either case it is a shame that Anandtech didn't do anything at all to discover where the disappointing results came from.
    (I hope nobody here actually thinks that ATI would go through the trouble of creating overdrive when it would only gain 0,5 fps)
    Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    this is getting way too geeky... Anand, you are indeed the master geek. I mean in this in a good way... or am I? hahaha hust kidding. You guys sure have a lot of patience. Incredible! Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    I'm just setting up a new rig with a 1/2in watercooling system. I have the maze4 GPU cooler and was about to buy a 9800-Pro. I assume that if I got an XT instead and used the C3.8 drivers it will probably run permenantly at the full overdrive speed because the chip should be cool even under full load?

    I wonder if the new drivers will cause any problems if I manualy overclock an XT...
    Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    lol, when ATI renders only the half in TRAOD it is clear that ATI is leading. Also please look at the 52.14 fps Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    Sounds like #28 has a case of the sour grapes ;) Reply
  • Anonymous User - Thursday, October 09, 2003 - link

    I think, reading the reference numbering that something has knocked them out of sync.

    Still, for want of a better reference, what I see as #12 has a good point. Try cooling the case and see what happens. I'd like to know how far the overdrive is prepared to take it.

    Also, perhaps the 9600XT part may see more dramatic improvements since the base running temperature should be lower than the 9800XT.

    Scott
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    OOOHHHH! that explains it! i didnt know the overdrive feature only went up to 432mhz...

    well, i guess its a step in the right direction, but still pretty conservative all things considered
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Checking out other sites reviews (I wonder why it wasn't mentioned here...), Overdrive only goes up to 432Mhz on the core and doesn't touch mem speed at all (although they said it will in future releases). That's such a small increase it's pretty much as if it wasn't there at all.
    Manual overclocking won't get replaced anytime soon.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Should try the 3.8 with IL2. My framerate went in the crapper and now there is texture tearing. They did fix the white water "perfect mode" bug though. Well,except in perfect mode it does about 20 FPS now. Whoopie..........! Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #33... was that a comeback? LOL
    Stop talking about HL2 cause it aint coming out for half a year now. Doom3 will most likely come out before it =)
    And why are some of you people acting like theres nothing to prove RIGHT NOW that ATi beats NV in DX9 games?
    Im not going to bash GeforceFX owners. You had your own reasons for buying that line of cards. Whether it was to support your favorite company or to get the really cool leaf blower. Who cares.

    But dont try and say no one knows if ATI cards will perform better then Nvidia. It does and its been proven. In developer interviews for almost all DX9 games, developers say they needed to code a special NV path to try and get compareable frame rates to the R3XX series. (With no special coding for ATI cards)
    If your going to tell me that the developers dont know, then whatever, put on your fanboy cape and jump of a bridge.

    p.s. ROFL @ #8
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #31, this is #28 ...

    I wasn't talking about HL2 frame rates.

    I was talking about TRAOD fram rates.

    I thuroughly expect HL2 to perform MUCH better than TRAOD. I'm really pissed off at CORE for writing a shitty game. I dont' think their performance problems are due to a lack of graphics power, but to incompitant programming.

    But I can play at 10x7 with AA and AF, as long as the horrid PS2.0 effects are off.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #31 atleast he can play the game...what are u using ? maybe intel extreme graphics... Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #31 was a reply to # 28. Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    are you telling me that your going to play HL2 , in DX8 at 640x800 on your Fx ? to get decent framerates ?


    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Nice of you to warn us ahead of time #8.
    It does take one to know one.
    So in your esteemed opinion, what are all the current GF FX card owners to do?
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    i am a little disappointed with the overdrive feature as well...i was all exicted too about the possibilities...and certainly expecting more than just a 1.5fps difference :(

    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #26 said: ""TombRaider AOD , the only DX9 game benchmarked is 3 1/2 times faster than Nvidia.
    ATI 42 fps , Nvidia 12 fps.

    WoooW.
    Imagine HL2.""

    Do0d, TRAOD has nothing to do with HL2. Also, NV is playable at lower resolutions than 16x12. And any card will fly with PS2.0 off...
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    ...I'm honestly struggling to work out why CM hyped these up as the best Catalysts ever. We knew these weren't going to be performance drivers, but the new features are gimmicky (overdrive/smartshader) or not as good as they should be (game profiles, but you have to manually switch them, bah!). Oh well, some bug fixes are at least nice.

    A somewhat disappointed 9800 Pro'er.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    TombRaider AOD , the only DX9 game benchmarked is 3 1/2 times faster than Nvidia.
    ATI 42 fps , Nvidia 12 fps.

    WoooW.
    Imagine HL2.

    I'm impressed, If Cat 3.8 bring more stability, that should be good enough for ATI owners.

    a Gforce2 mx owner.
    Reply
  • AgaBooga - Wednesday, October 08, 2003 - link

    #24, ATI would have had to implement special temperature measuring devices. I wonder how the bios flashing will respond to this... Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    sucks like 0.1% of gamers or market have the 9800TX why not do it with 9700pro or 9800pro


    just a though ...............




    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #20,

    You got your roman numerals backwards.
    VIII is 8
    IX is 9
    X is 10
    XI is 11
    XII is 12
    XIII is 13
    XIV is 14 and so on
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    SO, can anyone tell me what good is the 9800XT and Overdrive is good for now that the bundled game is not out and won't be out anytime soon? Yeah you can use it for benchmarks and some crappy DX9 games that are out right now but that's about it.

    By the time Half-Life 2 OR should I say HALF-LEAKED 2! arrives on the retail stores the new gen cards will be announced. Dates should be met because that is one of the factors we base our buying decisions.

    This is a big JOKE. Those big companies insult the gaming community with fiascos and goof-ups and treating us customers (fanboy or not) this way for the sake of money.


    REALLY I CAN'T WAIT FOR HALF-LEAKED 2 TO COME OUT!


    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Overdrive, great feature!
    AHAHAHAHAHAHAHAHAHAHAH
    Reply
  • WooDaddy - Wednesday, October 08, 2003 - link

    Hey Derek,

    Noticed a misprint. On the "Turning Up The Heat" page, it should be Final Fantasy 11 not 9. You got your roman numerals backwards; it's IX not XI. I almost got excited there. I'd love to see FF9 on PC. Shoot.. I'd love to see FF1-11 on PC..

    And before anyone responds.. NO I know there are emulators, but I don't do that stuff.. It's illegal unless I own it.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    The new stability properties/full uninstall and the new application profiles are great features.. However, I am very dissappointed with the overdrive feature.. It is the only feature that differentiates the new 9800XT and 9800Pro apart from the increased clock frequnecies, and it seems that the overdrive only increases the clock frequency another marginal 5% (412 vs 432 Mhz).. (But it translates a 15% increase in the clock frequency over a 9800Pro)


    About the discussion with Ps2.0:You can make two different conclusions: 1- If you want to get the best dx9 performance when the games will come out, wait for 6 months to see how Nv40 and R420 perform, then decide which one to pick (however, from the rumors I heard, do not expect that NV40 is going to be a total killer).. 2- If you want to buy the best card today that also can run ps2.0 games reasonably coming in the future, then ATI is the only way to go (and unfortunately TRAOD is the only game that is indicator to this, and Derek/Anand tried to avoid this conclusion as much as possible).

    And the possible performance improvements with the driver for Nvidia issue: It is very likely that this is biggest improvement thay can get.. After these ones, I only expect small improvements, not like an addition 30-40% again. So do not expect that NV3x series will beat R3x0 in close future as written in the benchmark article..

    Btw, I really appreciate what Derek/Anand did with IQ article.. It really seems a lot of work, but please this is not a motherboard/CPU article.. A gfx card article needs a lot of attention to image quality and because of the latest cheats, the images had to be analyzed very carefully.. I rather prefer to wait another week instead of a rushed benchmark.. Anandtech was a name for such quality before, please do not loose it because of your deals with IHVs..

    Thanks all..
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Better yet compare an overclocked 9800 Pro to a 9800 XT with Overdrive and with manually overclocked 9800 XT!

    This is what I want to know as a buyer. I dont care so much about STOCK speeds comarison, I care about the POTENTIAL of the card.

    Lastly I agree with #16 slightly lower resolutions than 1600x1200 should be tested since here is where big differences OCCUR. Or was it purposely done this way to show that there is a slight speed bump? in some way degrading the new ATI drivers.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    please don't divide the ffxi benchmark number by time taken; just give the raw score. i'd rather not have to do math in every review just to compare your numbers to my results. Reply
  • AgaBooga - Wednesday, October 08, 2003 - link

    I think that there should have been some tests at lower resolutions because at such a high resolution, you can't really gain a whole lot more just from drivers or a slight bump in speed. Reply
  • Locut0s - Wednesday, October 08, 2003 - link

    Another nice job guys! I don't mean to add more work to what must be an extreamly heavy work load, putting together that artcle in 1 day is impressive, but it would be nice if the charts also showed the min/max frame rates you got on these games. You already mension them in the text so why not put them up on the charts too. After all when you are talking about playing games at frame rates over 100fps it's the min frame rates that really start to matter, not aquezzing out another 0.5fps at the top. Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    #8,

    I'll gloss over the irony of calling yourself an idiot, except to say I agree. ;)

    When is it EVER a "good time" to buy a $400 card? Hardly ever.

    What if I'm in the market for a card NOW (because I'm still living with say a GeForce3, Radeon 8500, or lower for example), and I want to get the card that gives me the best possibility at being able to play the "upcoming" DX9 games?

    The Radeon is clearly the best bet.

    For all we know, BOTH of these cards will be crap with "upcoming DX9" titles. That doesn't mean that one of these cards clearly has the best potential.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Take note how he said that the highs and lows were closer together -might not look like much on the charts but might be a bit smoother in real life. Reply
  • sandorski - Wednesday, October 08, 2003 - link

    #3 has the best comment, is there a way to get the info of the Core/Mem speed being used? It just seems that outside of the OpenGL benches, the increases were too small to really know if Overclocking was happening or not. IOW, the gains were small enough to fall into statistical anomaly.

    Certainly any gain is good, but I was able to acheive better improvements from overclocking on a number of video cards, either ATI is being really conservative(which makes sense since they have Warranty concerns) or the Overdrive isn't working(possible, but I'd choose conservatism as the reason).

    I'd also be interested in the results of using exotic cooling(not that I'd use it). It would be especially interesting to see if the Overdrive feature is able to recognize Graphic Corruption or whether it is solely concerned with temperature.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    The XT is how the original 9800Pro should have been released 6 months ago Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    ZERO DX9 games???

    Then please explain how it is possible that I have a DX9 game on my machine at the moment. (A TWIMTBP even :-)

    But if the rare DX9 games at the moment are not the ones you are interested in, then indeed it is better to wait. (if your card is still fast enough for the current DX8 games).
    But if you do need to buy a new card at this moment, (maybe the old one broke) then the choice is very clear.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    the conclusion didn't say ati was a bad card to buy, it just said its a bad time to buy a card. Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    You're an idiot #8. HL2 won't be out for months and there are basically ZERO DX9 games out. Therefore, it'd be stupid to buy a card now based on basically no data. And no, ShaderMark 2.0 is barely hard data. Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Lol #4. ;)

    I am extremely perplexed myself with the conclusion. Specifically, the last sentence:

    "Even so, we are still standing behind our wait and see recommendation with respect to purchasing a card intended for use with the coming DX9 games."

    Hello? Anandtech? If you are buying these cards with an eye to DX9 performance, the choice is clear. ATI. At best, nVidia "optimizations" might get DX9 performance on par with ATI's. At worst, it's about 50% of ATI performance.

    It's only if you DON;'T care about DX9 performance that the FX line even becomes an option.

    Your conclusion is completley, 100% backwards.
    Reply
  • DerekWilson - Wednesday, October 08, 2003 - link

    I'd love nothing more than to strap a water cooling solution on this bad boy and watch it fly! Don't know if anyone's gonna let me tear stuff apart yet though ;-) Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Derek, Thanks for holding off on the buying advice Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    You should do another comparison that puts some extra cooling on the card, either by replacing the stock heatsink/fan or simply blowing more air across the card as if someone set a card cooler next to the video card. If the drivers worked well, this should noticeably improve card performance in a fairly inexpensive way. Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    Good benchmark.. but you know, you've probably struck a nerve with all the ATi fanboys with that conclusion ;)

    And yes, I own a Radeon. Hate of fanboys is (or should be) universal.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    I think you guys should've tried to find a way to see what it's actually overclocking to.
    .5fps is forgetable and considering the amount of cards that have been going over 460 core and 400 memory in old fashioned manual overclocking, it would be a nice reference point.
    Reply
  • Anonymous User - Wednesday, October 08, 2003 - link

    well done. Reply

Log in

Don't have an account? Sign up now